Paul Allen <paul.l.allen RemoveThis @NOSPAM.comcast.net> wrote in
news:20070718230948.07ab8f19@granite.localdomain:
> Apparently, you meant to say "larger pixels".
Yes. There's no second-chance edit capability in usenet, unfortunately,
and no way to guarantee that a correction will be read before the
original by others.
> I'm not sure why
> you would say the better signal/noise ratio of larger pixels gives
> no major improvement in image quality. Perhaps you've got some
> special meaning of "major" or "improvement" in mind?
I just can not find any examples of larger pixels or bins of pixels
giving any better results, if the data is properly handled. I believe
that the recent backlash against small pixels is caused by:
1) 100% pixel monitor viewing, and resampling routines that do not really
resample properly, but emphasize a subset of original pixels, thereby
decreasing the resolution without reducing the per-pixel noise.
2) #1 causing manufacturers of cameras and developers of RAW converters
to reduce noise excessively to cater to a training to look for noise at
100% or in poor downsamples; many of those noise reduction algorithms
only shift the noise to lower frequencies, and they do not disappear
readily when the images are downsampled (properly or not), or viewed from
a distance.
3) Poor philosophy - the idea that the noise of the captured pixel
directly determines the noise of the image.
>> What they need to do is stop damaging the output with heavy-handed
>> noise reduction.
> Shoot in raw mode and do the noise management yourself if it bothers
> you so much.
I do, but considering the fact that it takes 6 seconds to write the
extremely inefficient RAW files, I would like to use JPEG for quick
action, but the JPEGs are painfully dummied up in the fight against fine
noise.
> I'd rather have somewhat fewer clean pixels than lots
> of noisy ones.
I might, too, but it would depend on the trade-off. The fact is, the
quantum efficiency on the FZ50 sensor is above average (captures more
photons per square mm than many other cameras at a given sensor
illumination, including some big-pixel DSLRs), and read noise is better
than average, too, and lower than most Nikon DSLRs at all ISOs.
> In fact, if Panasonic were to introduce an FZ25
> combining the physical package and electronics of the FZ30 with the
> 5Mp sensor of the FZ20, I'd figure out a way to buy one.
Printed at the same size, with proper downsampling for the printer
resolution, I don't think that there would be any benefit over the FZ50.
What the FZ50 upgrades could really benefit from, IMO (assuming the
sensor size is kept small for lens design reasons), is some kind of
optimized read noise that leans a little more towards the higher ISOs.
Readout is optimal at ISO 200, and worst at 800 (worse than 1600).
Supposedly, the read noise can get down to the 1 electron range with the
reading of small photon counts. The way I've understood it is that CCDs
can have only one gain level at the first read stage on the sensor chip,
but the Nikon D40 seems to have readout optimized for higher ISOs, so
perhaps it is possible. But, of course, you will never collect a large
number of photons on a sensor that small at ISO 1600, so the highlights
will still be as noisy, even with a hypothetical zero read noise.
--
<>>< ><<> ><<> <>>< ><<> <>>< <>>< ><<>
John P Sheehy <JPS RemoveThis @no.komm>
><<> <>>< <>>< ><<> <>>< ><<> ><<> <>>< >> Stay informed about: Is Panasonic FZ50 being replaced 
FAQ
Search
Profile
Private Messages
Log in
